STONE, Circuit Judge.
This is an appeal from a decree declaring infringement of a patent.
The patent in suit is No. 1,326,908, issued to Samuel H. Boylan on January 6, 1920. Appellant made a device in accordance with a patent No. 1,484,695, issued to Charles E. Wood on February 26,1924. The master narrowly confined the claims of the Boylan patent and, as thus construed, found no infringement of that patent. Upon review, the court heard further evidence and, somewhat broadening the construction of the Boylan patent, found infringement thereof.
Estoppel, pleaded in the answer, is waived here. The validity of the Boylan patent is conceded. The contest is over the construction to be given to the claims of the Boylan patent. If the master is correct in his conception of the scope of that patent, there is no infringement. If the court is right in his conception thereof, there is infringement.
The device is purposed to separate pieces or particles of crushed zinc and lead ore bearing rock from a mass (called “pulp”) carried by water so that those of a similar size will he retained and thus classified. This is accomplished by placing the device in the path of a stream of water carrying the crushed ore so that the action of the water as it passes within and out of the device deposits therein pieces of the desired size, carrying the remainder beyond. The deviee is a suspended inverted metal cone with an outlet at the bottom controlled by a valve which is operated automatically by the weight of the contents. A clear statement of this purpose and its place in the process of separating ore from waste is contained in the report of the master and is as follows:
“The fundamental purpose of lead and zinc mills is to separate the lead and zinc mineral from the barren rock or waste. To do this, the mineral-bearing earth, called ore, is passed through crushers and ■ rolls until broken fine enough to pass through a screen of a certain size. It is then put on jigs which separate or classify the solid particles according to their specific gravity by pulsating water through the stream of pulp. This action causes the heavy minerals or lead and zinc to settle into a compartment from which they are drawn off, while the lighter particles or waste rock is carried off at’the top. The jigs are arranged successively, the first or rougher jig, making a concentrate of the lead and zinc mixed with some waste rock, and a tailing product supposedly free from mineral which goes to a waste pile. The concentrate goes to another jig called the cleaner jig, which in turn produces the lead and zine minerals free from waste material and a tailing product carrying more or less mineral. This tailing is ground finer to free the mineral from the waste and goes to a third jig which makes clean concentrate, and tailings still carrying some mineral. These tailings are then ground still finer and placed on concentrating tables. These tables are machines designed to concentrate finer material than is practical to handle on the jigs. They produce clean concentrate and tailings which go to the waste pile.
“In some mills the very fine slimes are removed at various stages in the process and treated with the oil floatation process. This process is particularly adapted to handle finer sixes than either the jigs or tables.
“In this process the ore is mixed with a considerable quantity of water and forms a pulp of water, mineral and waste and particles of sizes varying from impalpable dust called slime, to the largest size that will pass through the screen. The large amount of water and the extremely fine particles in this pulp are detrimental to some of the operations in the mill. A means of separating this fine material and the volume of water from the main portion of the material to be treated on the various machines was needed. Originally screens of various kinds were used, as also were boxes provided with an opening [50]*50in the bottom and an overflow spout at the top, into which the pulp was delivered. The coarser particles would settle and be drawn off with some water through the opening in the bottom, while the remainder of the water would overflow at the top carrying with it fine particles which would not settle. The screens were expensive to keep up and inefficient in separating the very fine sizes. The settling boxes were inefficient in removing all the water. For this purpose automatic cones found their field of use. They are used ahead of the jigs and also at the end of the jigs or any jSlace where it is desired to separate the coarser sized sand or mineral from the finer sizes and the greater part of the water.
“Machines of this type operate' on the physical principle that.the rate at which any particle will settle in water depends on its size and specific gravity, the coarser it is and the greater its specific gravity, the faster it will settle. For this purpose, only the size need be considered. These machines all have a cone or hopper which forms a receptacle in which this settling takes place. When the pulp is delivered into this receptacle, the coarser particles settle quickly and the finer particles, which do not settle so quickly, overflow with the water through the space provided. If the discharge of the coarser particles is restricted, it is readily seen that they will build up a bed in the bottom of the receptacle, thus increasing the weight of the total contents. If mechanism is applied to regulate the weight of the contents of this receptacle, it can be readily seen that this will regulate the depth of the coarser particles. This bed prevents the water from escaping with the coarser material to a very large degree, thus delivering this product almost entirely dewatered. When the depth of the bed is so great that only a shallow crater is left above it, the water introduced causes a more violent agitation and overflows more quickly, thus carrying out coarser sand than if the sand bed was more shallow and the crater deeper.
“Automatic maintenance of the relation between crater and bed will produce a uniform separation of the said particles according to their size, and this, together with the dewatering of the ore, is the purpose accomplished by such machines as plaintiff’s and defendant’s.”
The master has well described the device of plaintiff and its operation as follows:
“It consists of either a cone, or oblong inverted pyramid suspended from a single point by four rods anchored to the cone by rings. These rods are anchored at the upper end to a single fulcrum lever. From the long end of this lever there extends a bar diagonally downward to the bottom of the cone, which bar indirectly actuates a valve at the bottom of the cone or hopper. Further out on the long end of the lever is suspended a bucket into which is thrown iron or other heavy material, thus serving as a counter balance weight to the cone filled with water and mineral-bearing earth.”
“The ore mixed with water, is introduced at the top of the cone. The coarser particles settle to the bottom and form a bed of a depth determined by the amount of counter balance weight.
“The finer particles overflow with the water. When the weight in the cone becomes great enough, it descends, thus raising the counter balance weight at the end of the-lever and opening the valve. This permits the heavier and coarser materials to escape through the valve while the water carrying the finer particles of the mineral escapes through an outlet opening near the top of the cone. The size of the particles of mineral thus escaping with the water is regulated by the amount or volume of water in the crater of the cone, and the volume in its turn is regulated by the amount of the material in the bottom of the cone, which latter amount is in turn regulated by the counter balance weight. Should the material escape too rapidly through the valve, the cone becomes lighter, moves upward, the counter balance weights move downward, and the valve automatically closes until the level of material in the cone reaches the desired height. In this way, there is produced an automatic regulation of the level of the material in the cone, the volume of water in the eone and the size of the mineral particles escaping with the water.”
The court thought that there was infringement because the changes in defendant’s device were mere mechanical equivalents. Whether those differences were mere mechanical equivalents or were not, is the , question for solution on this appeal. To answer that question it is necessary successively to determine three matters: the scope of plaintiff’s patent; the differences in the two devices and the character, as to equivalence, of those differences.
I. The Scope of Plaintiff’s Patent.
The court and the master agreed that this patent covered not a basic or pioneer invention, but a combination. The contention of appellees seems to be that it was a [51]*51basic patent. He insists that the patents cited as anticipatory were, “with the possible exception of the Allen patent No. 1,147,356,” for “separators” while his device and patent are in another field being “classifiers.” The distinction, made by him, between “separattors” and “classifiers” is that separators are intended to and do only one thing, to wit, separate solids from a watery mixture, while classifiers separate solids of relatively the same size from the watery mixture. It is important to determine this character of the patent because thereon depends the liberality of construction to be given the claims and the resulting broadening or narrowing of the field of equivalents. Westinghouse v. Boyden Power Brake Co., 170 U. S. 537, 18 S. Ct. 707, 42 L. Ed. 1136.
A patent may be defined and limited by the terms upon which .it was allowed by the Patent Office and accepted by the patentee or by the state of the art. Here, both of these methods are urged by defendant.
(a) The Pile Wrapper.
As originally filed, the patent contained five claims. All of these were rejected — the first claim on Allen 754,911, in view of Allen 1,147,356, it being held that “there was no invention involved in placing the rod 7 above the tank A in the Allen patent 754,-911, in view of the later Allen patent”; claims 2, 3 and 4 were rejected for the same reasons on the same references; claim 5 on Allen 754,911 or Allen 754,732 or Gomez 1,-220,092 “in view of plate 6 of Caldecott, 1,-008,524.” Reference was cited also to Rapp 1,137,351 and Dull 1,005,522.
The first amendment expressly abandoned claim 5 which was very general and as follows:
“5. An automatic classifier of the character described, comprising a easing provided at its lower end with an outlet opening, a support plate arranged within the lower portion of the casing above and near the outlet opening, a valve to cover and uncover the outlet opening, and means operated by the movement of the casing to open the valve.”
Each of the other original claims was amended by inserting the words “and serving to entirely support the casing” as qualifying the “suspension means connecting the forward end of the lever with the easing.” The justification of the amended claims was stated as- being the simplicity of construction through “dispensing with any supporting frame or supports”.because the lever 17 “by being arranged above the casing 5, through the‘medium of the rods 11 to entirely support the casing.” It was stated that:
“The apparatus has a peculiar mode of operation, to wit, entirely suspending the casing 5 from the lever 17, and moving the lever by varying the weight of the casing, and utilizing this movement to open and close the valve 26. Applicant’s apparatus is characterized by its extreme simplicity.”
To escape the reference (Allen 754,911) it was stated that this Allen construction was “much more complicated” and “a frame work is necessary pivotally supporting the casing A at one end” and: “The easing slid-ably contacts with one end of the member 6, pivoted to the lever 7, which is pivoted to the frame.” To escape the reference Allen 1.147.356, it was stated: “The Allen patent No. 1,147,356, shows an apparatus having a wholly different mode of operation. The lever 22 pivoted above the casing A is connected with a float 6 for its movement. The casing A is stationary. This patent fails absolutely to teach the idea of entirely suspending the easing from a single lever arranged above it, and utilizing this lever to open and close a valve, at the lower end of the easing.” No mention was made of the other references cited by the Examiner.
These amended claims were rejected: “On Gomez, of record, it being held that there was no invention involved in suspending the casing from above in view of Allen 1.147.356, of record, Fig. 5. Allen shows the use of a suspending rod actuating a discharge valve in the same way as applicants. It appears that there is no invention involved in making the supporting rod in Gomez, a suspending rod in view of Allen. See also Allen, 942,697, Dec. 7, 1909 (83-8240), Fig. 6.”
A second time the claims were amended to declare the lever and valve to be attached to, supported by and moving in relation to the casing. The justification of these amended claims was that:
“Applicant has produced a novel combination wherein a single pivoted lever is arranged above the vertically movable easing, and connected therewith to entirely support the¡ same. The lever 27 is pivoted to the lower end of the casing, and is supported thereby. This lever carries the valve 26, which is movable with relation to the easing S. The lever 17 is connected with the lever 27, by the link 31, By this organization no complicated supported frame is employed, nor a complicated connection between the lever 17 and lever 27. Compare this simpli[52]*52fied construction with the complicated structure of Figure 7 in patent No. 1,220,092. See the lever 20a pivoted to frame 9a, roller 21 carried by lever 20, and contracting with cam 38 of lever 30. Obviously applicant dispenses with a large number of elements, without dispensing with their functions.
“The Allen patent No. 1,147,356, Figure 5, shows a stationary casing A2, and a float 6a, carried by lever 30 and connected with valve. This patent fails to teach the idea of operating the valve by the vertical movement of the easing A2.”
These second amended claims were rejected: “On Gomez, Rapp, or Allen, 942,697, of record Fig. 6, as devoid of invention in rising a single lever above the casing in view of Pooley et al., 125,612, April 9, 1872 (73-175); Pierce, 613,866, November 8,1898 (73-175); Bradley, 406,254, July 2, 1889 (73-175); McNeill, 1,112,636, October 6, 1914 (73-177).”
Reconsideration was asked on the grounds that: “In connection with the patents cited in the last office action, it is pointed out that such patents disclose automatic weighing machines, which embody a wholly different principle of operation. As understood, those machines have means to dump the load after a certain amount of material is fed into the easing or pan, but there is no means to maintain the weight of the load constant, as in applicant’s device. Furthermore, the claims are not anticipated, since the combinations of the same are not present in any one reference.”
Reconsideration was denied, the Examiner stating:
“It is not clear to the examiner what differences in structure as well as what differences in function applicant ascribes to his claimed devices over the references last cited. —It appears that both in applicant’s case as claimed, as in the references, the material is discharged after a given amount has accumulated. Unless applicant can distinctly show patentable substance over the prior cited art in the present claims or in such claims as he may present in his next action, the next action by this office will constitute a final by the examiner action in this case. The claims stand rejected.”
A third amendment was presented in which claims 1 and 2 were canceled; claims 3 and 4 renumbered as 1 and 2 and a new claim 3. In support of these amended claims, the applicant stated:
“In classification of mineral-bearing sand, the primary object is to separate, through the agency of water, the coarse from the fine solids which go to make up the bulk of the feed. It is therefore obvious that every successful device for the separation of the coarse from the fine solids must be automatic in its operation both as to the overflow carrying away the fines, and the discharge of the coarser material through the valve at the base of the conical easing. It is also understood, although not as obvious except on dose study of the hydraulic principle, that the formation and depth of the crater formed by the discharge of the feed in the center of the conical casing, together with the natural force and effect of the water used to move the feed and which is given a boiling action by its fall, is of primary importance in effecting a true classification, since the hydraulic action frees and raises all of the relatively fine particles which pass off in the overflow. To effect and maintain the crater at its highest efficiency, which is a practically constant depth of crater, the device must necessarily operate automatically and be exceedingly accurate and sensitive to the varying conditions of the feed. The device must of necessity eliminate friction by the elimination of all unnecessary' levers, rollers, guides or anything that could possibly cause friction which would interfere with its free action. Applicant’s device is entirely suspended upon a knife blade edge, the only possible friction lying at the pivotal connection of the rods. These pivotal connections have been reduced in number as far as possible, to give the easing free and sensitive vertical movement. The efficient operation of the device renders it necessary to maintain as nearly as possible a certain amount of sand in the easing forming the crater having a positive depth. If too large an amount of sand is allowed to accumulate in the easing the crater fills up and becomes shallower which forces a portion of the coarser material into the overflow causing inefficient classification. If too much of the coarser sand is removed through the valve at the base of the casing at one time the crater formed in the sand deepens and allows finer particles of the same to go down with the coarser particles which also impairs the classification. In applicant’s device these difficulties have been overcome by the elimination of friction to the free and sensitive movement vertically of the casing, which has been shown under operation to maintain an almost constant depth of crater in the sand within the casing.
“The patent to Allen, No. 1,147,356, cited, shows not a classifier of solids, but a settler and separator for solids and liquids in the treatment of pulp and slimes. The in[53]*53ventor lays no claim to its being a practical classifier; it has a stationary cone or hopper and a float within the hopper to operate the valve at the base of the hopper. The Examiner will readily see that sand binding around the valve stem (Fig. 1) and valve will have a tendency to prevent lifting the valve from its seat and after it is lifted would-prevent its return. The amount of friction here eliminates this device as an efficient classifier.
“The patent to Rapp cited is subject to the same objection. While the valve does not open in the easing the actuating mechanism therefor passes through the casing and the sand. Furthermore, Rapp shows the valve stem b2, as supported through a guide b3. This guide would be subject to the action of fine particles of sand and consequent friction. It is also thought that Rapp’s device as shown is inoperative in that if the easing al tilts there would be a binding action at the upper head of the valve stem b2 preventing the operation of the scale beam.
“Gomez shows a complicated system of levers, rollers and pivots which aim at applicant's principle but do not accomplish it, Gomez finding it necessary to provide four pivots and a roller to accomplish his result. Furthermore, the upper end of the easing must be supplied with the guide against which the casing is forced in its upward movement. This device is not practical as a classifier of solids and is not claimed as such by the inventor. It retards the movement of the water to allow the matter held in solution to settle in the apparatus and the clear water to flow out. * *■ * The sediment is discharged automatically from time to time. It suggests separation of fluids from solids, but not classification of solids in any practical sense.
“In the claims now presented applicant recites the use of a single vertically swinging lever supporting a casing, a second pivoted lever carrying a valve and means connecting the pivoted lever and the valve lying entirely outside of the easing. It will be seen that applicant uses but two pivots and a balance and that the parts are so disposed as to prevent their contacting with frictional faces.
“New claim 3 further recites the adjusta-bility of the casing and means for adjustably connecting the balance lever and the valve not shown by the references.”
These claims were allowed.
Consideration of the progress of this case through the Patent Office, as revealed by this file wrapper, make clear the end sought by plaintiff, the problems he deemed present and his method of solving them through his device. The end sought was to separate, from a water mixture, zinc or lead ore bearing rock and collect solid particles of approximately the same size. The action of this mixture in an inverted cone tended, through gravity, to precipitate particles to the bottom. If the level of solid matter thus deposited was kept at a proper height in the cone, the particles deposited would be near enough the same size for practical purposes. If such height were not fairly well maintained this result would not follow. Therefore, the problem was to keep this deposit constantly as near the desired height as possible. In operation, the mixture was continuously flowing into and out of the cone making such deposit as the conditions there permitted. Hence, the solution of the above problem depended on a decrease of the solid matter already deposited in relative harmony with the new matter being constantly deposited from the inflowing mixture. The outlet for deposited matter was at the bottom of the cone and was controlled by a valve which closed that opening. Thus, the problem further reduced itself into devising such control of this valve that the outlet of deposited matter would correspond to the new matter being continuously deposited and thus the level of the deposited matter be maintained. Obviously, this required automatic action of the valve in releasing or restraining the outpour. Obviously, this automatic action of the valve must be directly regulated by the new deposits. This synchronized valve action could be secured by having the valve controlled by the weight of the cone as affected by the new matter. Therefore, the solution of the problem was offered through such control of the valve by the varying weight of the loaded cone, thus the automatic feature was provided. It is also clear that the more readily or sensitively responsive this control could be made, the more nearly constant would the level in the cone be maintained. Therefore, the problem became, in its last analysis, a search for a device which would “eliminate friction by the elimination of all unnecessary levers, rollers, guides or anything that could possibly cause friction which would interfere with its free action” (plaintiff’s last letter to the Examiner). The method he followed in building a device with this minimum of friction or interference is stated by plaintiff in his last letter to the Examiner as follows: to suspend the device “entirely * * * upon a knife blade edge, the only possible friction lying at the pivotal connection of [54]*54the rods”;- to reduce the pivotal connections as far as possible so as “to give the easing [cone] free and sensitive vertical movement” ; to place the mechanism entirely outside the cone so as to escape interference from the material therein; to reduce to the minimum levers, rollers, pivots and other parts having frictional contacts. Plaintiff sums up his accomplishment as follows:
“In the' claims now presented applicant recites the use of a single vertically swinging lever supporting a easing, a second pivoted lever carrying a valve and means connecting the pivoted lever and the valve lying entirely outside of the casing. It will be seen that applicant uses but two pivots and a balance and that the parts áre so disposed as to prevent their contacting with frictional faces.”
Thus, the file wrapper makes clear that what plaintiff claimed, what he was willing to and did accept and only what the Examiner would allow was a particular combination of parts in a device which rested, for patentability, upon the proposition that such combination reduced friction or interference with the outlet valve to the minimum and thus more perfectly accomplished the result of maintaining the level of solid matter in the cone.
(b) The Prior Art.
The citations from the prior art introduced by defendant to limit the scope of the patent are, in chronological order, as follows: Reinert 641,227 (January 9, 1900), Allen 754,732 (March 15, 1904), Allen 754,-911 (March 15, 1904), Dull, 1,005,522 (October 10, 1911), Allen 1,147,356 (July 20, 1915), Gomez 1,220,092 (March 20, 1917) • and Dull 1,258,100 (March 5,1918). Inspection of the claims and devices covered by these patents reveals much that appears in plaintiff’s device. One argument advanced by plaintiff is that most, if not all of these devices are for “separators” and not for “classifiers.” This argument is not well founded as to any of these patents because several of them are for “classifiers” while the remainder are in the same or, at very least, a closely allied art and, therefore, are properly citable as suggestive to one attempting to devise a classifier. All of them are devices to separate solid matter from a watery mixture; all of them employ an inverted cone in which to collect the separated solids; all have an opening at the bottom of the cone to release the solids; all have a valve controlling this outlet; all govern the operation of this valve automatically by the weight of the loaded cone; in all, the valve tripped when the loaded cone reached a determined weight and closed when that weight was reduced below the determined standard. Thus, the broad purpose of separating solids from a watery mixture through the employment of an inverted cone and the release of the collected solids through a bot.tom outlet controlled by a valve which was governed by the weight of the loaded cone was well known and established when plaintiff entered the field. But plaintiff had in view a different and narrower purpose than separating solids from the mixture. His purpose was to separate solids of relatively the same size from the mixture. T.he Allen 754,732 was not designed to and could not be a classifier because the device consisted of two superimposed cones connected by a tubing the lower cone holding the solids, thus the mixture did not directly enter the collecting cone and there was absent the essential conditions 'of a classifier, which were a constant level of solids in the cone and the direct discharge of the mixture into that loaded cone. In all of the other citations, there is but one receptacle (a cone in all except Dull 1,005,522) into which the mixture pours. Operation of all of these, except Allen 1,147,356, is governed by the weight of the loaded cone (apellee’s method), which weight corresponds to the level of solids and in each the weight or level is subject to regulation. Therefore, no reason has been shown why any of these (except possibly Dull 1,-005,522 and Allen 1,147,356) could not be made to act as a classifier. In all of them, except Allen 1,147,356, the valve is actuated by downward vertical movement of the loaded cone. In Reinert, the cone is rigidly attached near the bottom to a counterweighted lever and tilts when beyond the tripping weight. In Allen 754,732 and Allen 754,911, the cone is pivoted at'one side of the top and tilts to the overload. In Dull 1,005,522, the cone is supported at one side on knife-edge trunnions resting in shallow bearings and tilts thereon to the load. In Gomez 1,220,-092, the cone is supported near the bottom by a counterweight lever and settles, vertically between guiding standards, with the overweight. In Dull 1,258,100, the cone is suspended on two rods at opposite sides and moves downward under weight (the operation of the valve is somewhat different in this patent from that of plaintiff or the other citations). In several (Allen 754,732, Allen 754,911, Dull 1,005,522 and Gomez 1,220,-092) ■ the valve connections are entirely outside the cone. In the others (Reinert, Allen 1,147,356 and Dull 1,258,100), they are [55]*55within the eone. Without collecting or detailing other features of these citations, it is clear that there had been much investigation and discovery along the same lines, with the same principles of operation and very similar (often identical) features of construction as employed by appellee.
Therefore, it is clear that appellee’s device was secondary, was for a combination and was for a combination which must be very closely construed to its exact members, relations and operation, if it is to avoid the prior art and to remain within the accepted meaning as revealed in the file wrapper.
Infringement.
In the light of this close construction of appellee’s patent and the resulting narrow range of equivalents, the matter of infringement rests upon a comparison of the two devices involved here. The cardinal distinguishing characteristic claimed by appellee for his device is its simplicity and accuracy of operation. He says he attains this by eliminating friction of operation of the vaxve through his method of suspending the cone from a single knife edge, placing all operating parts outside of the cone and decreasing the number and friction contacts of those parts. Placing the parts outside the cone was old and introduced ni> new element or feature. He has no fewer valve operating parts than Allen 754,732 or Dull 1,005,522, both of which were outside the cone. The method of supporting the cone is very different from all of the citations except Dull 1,258,100. There is some general similarity in principle to that patent of Dull, but there is a difference in flexibility and number of supporting parts which would naturally give greater freedom of movement and less interference, through friction, in favor of appellee’s device. There are some features of the construction of the interior of the eone which are new, to wit, the vertical ring 6 in the upper part of the cone and the plate 34. near the bottom. Only the first of these is covered in the claims (Claim 2) and neither is urged here.
Compared with appellant’s device, as to these cardinal features of suspension of the cone and number of valve operating parts, there appears quite a difference. As to the suspension, the difference has been well likened, by counsel, to that between a yard arm scale and a platform scale. Boylan’s entire device is suspended from a single horizontal lever resting medially upon a single knife-edge support. From one end of this lever, the cone freely hangs; to the other end is movably attached the long lever which extends downward to the outward end of the valve lever. The counterweight, which controls the action of the cone is placed on the supporting lever toward the end opposite that to which the cone is attached. Appellant’s suspension is made up of three connected units — the cone, the rigid supports and the member connecting those two. Appellant’s eone is rigidly attached to a square frame placed somewhat below the upper edge of the cone. Two opposite sides of this frame are extended at both ends with knife-edge bearings on the under sides of these projections crosswise at the ends thereof. The rigid support is of a separate unit which does not come in direct contact with the cone or cone frame. It is a rigid frame placed on the ground and having two elevated horizontal members paralleling the extended sides of the eone frame. Between ,these horizontal members two other oppositely horizontal rods are rigidly attached on the upper sides of which near each end are welded or attached upwardly projecting knife-edge bearings. The connecting members are as follows: two pairs of horizontal rocker arms having holes near their outer ends through which pass the above cross supports of the rigid supporting member in such wise that the upper surface of the holes rests upon the four upwardly projecting knife-edge bearings; the outer ends of the two paralleling arms rigidly attached to a connecting rod which parallels on the outer side the supports having the upper knife-edge bearings and which also has knife-edge bearings on the upper sides near the ends; one pair of the rocker arms are hinged at the inner ends on a bolt to which is movably attached a lever extending downward to connect with the valve lever; the other pair of arms are not joined but have holes near the inner ends from which are suspended the counterweights governing the movement- of the cone. The eone frame is suspended on four C shaped clevises — the upper inside surface of the clevises resting on the knife-edge bearings on the rods connecting the rocker arms and the lower inside surface of clevises supporting the knife-edges on the projecting ends of the cone frame. From the above description of the two devices, it is seen that the method of support is very different in the two devices. In Boylan, there are very few suspension parts and the entire suspension is from above. In Wood, the entire suspension is from below and is made up of three distinct units, each, in turn, composed of several parts. Also, Boylan’s cone may vary its posi[56]*56tion horizontally rather freely while Wood’s cone has practically no lateral motion. Boy-lan’s may tilt, Wood’s cannot.
As to valve operating parts, consideration of the two devices will at once suggest the wide difference between the very few used by Boylan and the many used by Wood. Beside the one supporting knife-edge, Boylan has but four points of frictional movement. Wood has fifteen, including twelve knife-edge contacts.
About the only similarities between the two are that each releases the solid matter by a valve controlled by the varying weight of the loaded cone influenced by counterweights — all of which were old in the art.
We think the decree should be reversed with instructions to set the decree aside and enter decree for defendants below. It is so ordered.